Clinicopathological characteristics and prognostic factors for intrahepatic cholangiocarcinoma: a population-based study

We aimed to explore the clinicopathological features and survival-related factors for intrahepatic cholangiocarcinoma (ICC). Eligible data were extracted from the Surveillance, Epidemiology and End Results (SEER) database from 2004 to 2015. Totally, 4595 ICC patients were collected with a male to female ratio of nearly 1:1. The higher proportion of ICC patients was elderly, tumor size ≥ 5 cm and advanced AJCC stage. Most patients (79.2%) have no surgery, while low proportion of patients receiving radiotherapy (15.1%). The median survival was 7.0 months (range 0–153 months). The 5-year CSS and OS rates were 8.96% and 7.90%. Multivariate analysis found that elderly age (aged ≥ 65 years old), male, diagnosis at 2008–2011, higher grade, tumor size ≥ 5 cm, and advanced AJCC stage were independent factors for poorer prognosis; while API/AI (American Indian/AK Native, Asian/Pacific Islander) race, married, chemotherapy, surgery and radiotherapy were independent favorable factors in both CSS and OS. Furthermore, stratified analysis found that chemotherapy and radiotherapy improved CSS and OS in patients without surgery. Age, sex, race, years of diagnosis, married status, grade, tumor size, AJCC stage, surgery, chemotherapy and radiotherapy were significantly related to prognosis of ICC. Chemotherapy and radiotherapy could significantly improve survival in patients without surgery.

Covariates and endpoint. Patient features were analyzed according to relevant factors: age (˂ 65, ≥ 65); sex (female, male); race (black, white or API/AI); marital status (unmarried, married); insured status (uninsured/unknown, any medicaid/insured); year of diagnosis (2004-2007, 2008-2011, 2012-2015); grade (grade I/II, grade III/IV, unknown); tumor size (˂ 5 cm, ≥ 5 cm, unknown); 6th edition of AJCC stage (stage I, II, III, IV); surgery (no surgery, local tumor excision/segmental resection, lobectomy/hepatectomy), chemotherapy (no/unknown, yes), radiotherapy (no/unknown, yes). To be specific, unmarried population included divorced or separated, single (never married or having a domestic partner) and widowed 13 . Year of diagnosis was equally divided which was referred to a previous study 14 . The stratification of age and tumor size was also based on previous researches 15,16 . API/AI means American Indian/AK Native, Asian/Pacific Islander. In addition, the staging of cancer is based on the 6th edition of AJCC stage system, which adapted to patients in the SEER database with a diagnosis time of 2004-2015.
Overall survival (OS) and cancer-specific survival (CSS) were taken as the study endpoint. OS was defined as the interval from diagnosis to all-cause death, while CSS referred to the interval from diagnosis to ICC-caused death. The cut-off date was set on November 31, 2018 because it was pre-determined until November 2018 (with death data) in accordance with SEER 2018 submission database.
Statistical analyses. Univariate analysis was estimated by Kaplan-Meier (K-M) method, followed by assessment of the differences of CSS and OS using log-rank test. Parameters with P value ≤ 0.2 in univariate analysis were further evaluated in multivariate Cox proportional hazard model 17 . Stratified Cox regression model was conducted, aiming at assessing the prognostic effects of chemotherapy and radiation in different subgroups stratified by surgery style. SPSS software (SPSS Inc., Chicago, USA, version 19.0) was employed for statistical analysis, and GraphPad Prism 5 was utilized for plotting survival curve. A two-sided P < 0.05 indicated statistical significance.

Results
Patients' features. There were 8953 ICC patients from 2004-2015 totally, and the number of patients was increased year by year ( Fig. 1). According to the exclusion criteria, 4595 patients were enrolled after screening. The specific screening process was shown in Fig. 2, and features of patients as well as therapeutic regimens were shown in Table 1 Tables 3  and 4, compared to the non-chemotherapy group, chemotherapy group was associated with better CSS and OS in patients who did not receive any cancer-directed surgery (P < 0.001). But for patients with surgery did not show significant survival benefit (Table 3). In the stratified analysis of non-radiation group and radiotherapy group, similar results were obtained. Patients in the no surgery group received significant survival benefits after radiotherapy (P < 0.001), whether CSS or OS, while patients in the surgery group did not (Table 4).  www.nature.com/scientificreports/

Discussion
ICC is a subtype of bile duct adenocarcinoma involving liver small ducts 18 , and the second most common primary liver malignancy after HCC 19 . Due to its rarity, few large-scale researches are available for instructive conclusions on proper management for ICC patients 20 . For this purpose, we included a total of 4595 ICC patients to investigate the clinicopathological features and to examine survival-related factors of ICC. The incidence of ICC has been increased in the US in the last forty years (1973-2012), from 0.44 to 1.18 cases per 100,000 21 , and its incidence is also increasing throughout the world 22 . Previous studies report that ICC patients are elderly, without clear sex differences 23 , which are consistent with our study. Besides, we found that a large proportion of ICC patients had tumor size ≥ 5 cm and advanced AJCC stage. The outcome of ICC is  24 . Despite hepatolithiasis, viral hepatitis B and C, cirrhosis and primary sclerosing cholangitis reported as risk factors by various researches, data from Eastern and Western countries are not identical [25][26][27] . Apart from AJCC staging and histological grade, tumor size ≥ 5 cm 24 and marital status 14 have also been found to be significant prognostic factors for ICC. Additionally, we found that age, sex and race were also important prognostic factors.
Radical surgery is the only curative treatment, including major liver resection with extended systematic lymph node (LN) dissection 28 , which is recommended by most institutes. However, the resectable rate of ICC is still low, varying from 19 to 74% globally 29 . In our study, only 20.9% of patients underwent surgical treatment. Unresectable ICC patients are generally treated by systemic chemotherapy. ABC-02 trial revealed significant survival advantage in patients with advanced biliary cancer who were treated by gemcitabine/cisplatin combined chemotherapy than those with gemcitabine alone. Other combined regimens included gemcitabine-or fluorouracil-based chemotherapy 6 . NCCN guidelines recommend radiation for subjects with positive regional LN or microscopic tumor margins (R1) following cancer-directed resection 30,31 . And our research found that significant survival benefits of radiation and chemotherapy in non-surgery group according to stratified Cox model (P < 0.0001), which were consistent to previous studies 32, 33 .
With using advanced technologies like next-generation sequencing (NGS) in ICC, recent research starts to reveal the genetic and molecular processes behind carcinogenesis. The results concluded through empirically studying the genome profiling, epidemiology and experiments offer novel insights into genomic formation, risk factors, cellular origins and constructing tumor microenvironment to the pathogeny of ICC. As a recent www.nature.com/scientificreports/ retrospective study verifies, the treatment with blockage of Her-2/neu in ICC patients suffering gene amplification has great potential 34 . Immunotherapeutic progress can also offer new opportunities for ICC therapy 35 . After PD-1 inhibitor treatment, a complete response was founded in the chemotherapy refractory metastatic ICC patient who suffers mismatch-repair deficiency (dMMR) 36 . Unfortunately, there is no information on molecular genetic profiles and targeted therapy in the SEER database. SEER database is the largest publicly accessible and authoritative source on cancer incidence and survival. Therefore, our findings could guide clinical management by using the large-scale, reliable research dataset. As far as we know, our study is largest population-based one to detect prognostic indicators in ICC. Inevitably, there are also several limitations in our study. Firstly, due to the nonrandomized nature of this study, selection bias is inevitable 9,11 . Secondly, certain important factors, including tumor gross type, depth of invasion, status of harvested lymph node, molecular-genetic profiles, metabolic abnormalities of liver and chronic liver disease (viral infection and cirrhosis), were inaccessible in SEER dataset. Thirdly, detailed data on chemotherapy and radiotherapy were not available. Although it is better to obtain more details, we believed that the present available data from SEER database could fit our research objectives very well. Further studies should investigate the above concerns.

Conclusions
In the present study, we investigated the clinicopathological features and survival of ICC patients. Age, sex, years of diagnosis, grade, tumor size, race, AJCC stage, married status, surgery, chemotherapy and radiotherapy were significantly associated with prognosis. For patients without surgery, chemotherapy and radiotherapy showed significant benefits to improve survival. Hopefully, our findings are of great significance for clinical management and future prospective studies for ICC.